CN115424145A - Planetary remote sensing image surface terrain change detection method based on deep learning - Google Patents

Planetary remote sensing image surface terrain change detection method based on deep learning Download PDF

Info

Publication number
CN115424145A
CN115424145A CN202211136723.5A CN202211136723A CN115424145A CN 115424145 A CN115424145 A CN 115424145A CN 202211136723 A CN202211136723 A CN 202211136723A CN 115424145 A CN115424145 A CN 115424145A
Authority
CN
China
Prior art keywords
remote sensing
sensing image
planet
feature
change
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202211136723.5A
Other languages
Chinese (zh)
Inventor
戴育岐
郑铁
薛长斌
周莉
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
National Space Science Center of CAS
Original Assignee
National Space Science Center of CAS
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by National Space Science Center of CAS filed Critical National Space Science Center of CAS
Priority to CN202211136723.5A priority Critical patent/CN115424145A/en
Publication of CN115424145A publication Critical patent/CN115424145A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/10—Terrestrial scenes
    • G06V20/13—Satellite images
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
    • G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
    • G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
    • G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Computing Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Multimedia (AREA)
  • Artificial Intelligence (AREA)
  • Software Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Databases & Information Systems (AREA)
  • Medical Informatics (AREA)
  • Biomedical Technology (AREA)
  • Mathematical Physics (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Astronomy & Astrophysics (AREA)
  • Remote Sensing (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)

Abstract

The invention belongs to the technical field of remote sensing image processing and change detection, and particularly relates to a planetary remote sensing image earth surface terrain change detection method based on deep learning, which comprises the following steps: inputting the acquired double-time-phase planetary remote sensing image data of the area to be detected into a planetary remote sensing image change model which is established and trained in advance, and obtaining a surface terrain change detection result; the planet remote sensing image change model extracts multilayer features of double time phase images based on a twin network framework, and comprehensively measures the difference features of different levels through a multi-level difference feature fusion structure to realize detection of terrain change. The planet remote sensing image change model designed by the invention not only has high accuracy, but also has the advantages of less parameters and low calculation complexity, and has higher practical application value.

Description

Planetary remote sensing image surface terrain change detection method based on deep learning
Technical Field
The invention belongs to the technical field of remote sensing image processing and change detection, and particularly relates to a planetary remote sensing image earth surface terrain change detection method based on deep learning.
Background
The change detection is the characteristic and process of quantitative analysis and determination of surface change from the remote sensing data in different periods. The change detection of the remote sensing image is essentially to detect the change information of the earth surface in different time phases, and is to quantitatively analyze and determine the characteristics and the process of the earth surface change from the remote sensing data in different periods. The Mars detector which has been transmitted in China returns a large amount of image data when executing a scientific detection task, the significance of analyzing the data collected by the planetary detection task by using a change detection algorithm is important, the mining and extraction of value information in a time dimension are mainly used, the application scene of the change detection in the process of the planetary detection at present mainly comprises the observation of the change of the planet landform caused by meteorite impact and the observation of the change of the slope striae landform on the surface of the Mars, and the important significance is realized in the aspects of obtaining the dynamic change information of a key attention area, adjusting a response scheme in time and the like. In addition, when the satellite is in a staring observation state, namely the visual angle posture of the camera is not changed, and a certain target area is continuously photographed at the same position, repeated images can be removed and effective information can be extracted by detecting the on-orbit change of the sequence images, so that the required downloaded data volume is reduced to a greater extent, and the bandwidth pressure is relieved.
With the popularization of high-resolution remote sensing images, change detection has become one of the most studied subjects in the remote sensing field. The traditional image change detection method comprises the steps of image processing, difference map generation, difference map analysis and the like, the mainstream methods in the difference map generation stage comprise a difference method, a ratio method, wavelet fusion, a logarithmic ratio method and some related improved operators, the methods mostly need domain prior knowledge, then change detection is carried out by manually designing and extracting the characteristics of texture, morphology, neighborhood and the like of an image, the extracted image characteristics are shallow, the image is difficult to completely model, and time and labor are consumed. In recent years, a change detection method based on deep learning is developed and gradually becomes a preferred tool for remote sensing image analysis. The application of deep learning in the remote sensing image change detection has ideal effects on solving the limitation of an image processing algorithm, improving the precision of the change detection and the automation degree of the process.
On the basis of the deep learning method, a set of more intelligent planetary remote sensing image change detection process is designed by building a network and defining optimization rules, so that the method has important research value and theoretical significance for further improving the change detection precision and better giving full play to the potential application value of a deep learning model.
In recent years, many organizations and scholars at home and abroad explore the technology for detecting the change of the planet remote sensing image, most of the organizations and scholars adopt a solution scheme of dividing a plurality of sub-problems into a plurality of sub-problems and treating the sub-problems by means of feature extraction, post-classification or difference map generation, analysis decision making and the like, and the defects of complicated steps and error accumulation still exist.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides a novel planetary remote sensing image change detection model and method based on deep learning so as to obtain a faster execution speed and a better detection accuracy result. The new change detection algorithm is a scheme capable of directly obtaining an expected change detection output result from original double-time-phase image data, and the whole change detection model mainly comprises a double-time-phase image feature extraction module, a multi-level feature difference fusion module and a classification decision module. In the training process of the change detection model, artificial subproblem division is not carried out, and an end-to-end learning mode is adopted, so that a global optimal solution can be obtained more possibly.
In order to achieve the purpose, the invention provides a method for detecting the surface topography change of a planet remote sensing image based on deep learning, which comprises the following steps:
inputting the acquired double-time-phase planet remote sensing image data of the area to be detected into a planet remote sensing image change model which is established and trained in advance to obtain a surface topography change detection result;
the planet remote sensing image change model extracts multilayer features of double time phase images based on a twin network framework, comprehensively measures the difference features of different levels through a multi-level difference feature fusion structure, and realizes detection of terrain change.
As an improvement of the method, the input of the planet remote sensing image change model is remote sensing image data after the planet is registered in the same region to be detected at the time T1 and the time T2, the type of the remote sensing image data comprises a visible light image and a multispectral image, and the remote sensing image data is output as a detection result of whether the surface topography changes; the planet remote sensing image change model comprises a double-time-phase image feature extraction module, a multi-stage feature difference fusion module and a classification decision module; wherein the content of the first and second substances,
the double-temporal image feature extraction module is used for extracting multilayer features of double-temporal images and enhancing the local and global visual feature characterization capability;
the multi-level feature difference fusion module is used for comprehensively measuring the difference features of different levels and enhancing the feature expression capability;
and the classification decision module is used for outputting a detection result, wherein 0 indicates no change, and 1 indicates a change.
As an improvement of the method, the double-temporal image feature extraction module adopts a basic twin network framework and comprises two improved MobileViT networks with the same structure, the improved MobileViT networks take a MobileViT backbone network part as a feature extractor, a final pooling layer and a classifier of an original MobileViT are removed, and three feature positions are respectively arranged at the output of the mv 1 module, the output of the mv 2 module and the output of the improved MobileViT network of each improved MobileViT network, so that a first feature Stage1_ T1, a second feature Stage2_ T1 and a third feature Stage3_ T1 of the planetary remote sensing image data to be detected at the time of T1, and a first feature Stage1_ T2, a second feature Stage2_ T2 and a third feature Stage3_ T2 of the planetary remote sensing image data to be detected at the time of T2 are obtained.
As an improvement of the above method, the processing procedure of the multi-level feature difference fusion module MFDF includes:
respectively calculating absolute difference values of the features Stage1_ T1 and Stage1_ T2, stage2_ T1 and Stage2_ T2 and Stage3_ T1 and Stage3_ T2 to obtain corresponding first-Stage absolute difference value Diff 1 Second order Diff 2 Third order absolute difference Diff 3 ；
Figure BDA0003852409590000031
Diff is mixed 1 And Diff 2 Performing cascade operation to obtain a first cascade output Fdf 1 And then Fdf 1 Sequentially inputting the 3 x 3 standard convolution layer and the maximum pooling layer to obtain a second cascade output Fdf 2 And Fdf 2 And Diff 3 Cascade to obtain the third cascade output Fdf 3 And output to the classification decision module.
As an improvement of the method, the classification decision module comprises an average pooling layer and a full connection layer.
As an improvement of the method, the method further comprises a training step of the planet remote sensing image change model; the method specifically comprises the following steps:
establishing a training set by using the public Mars image data;
inputting sample data in a training set into the planetary remote sensing image change model in batches, calculating a cross entropy loss function, performing gradient back propagation, updating network weights by adopting an Adam optimizer, and performing iterative updating until the training requirements are met to obtain the trained planetary remote sensing image change model.
On the other hand, the invention provides a system for detecting the surface topography change of a planet remote sensing image based on deep learning, which comprises the following components: detecting an output module and a planet remote sensing image change model; wherein the content of the first and second substances,
the detection output module is used for inputting the acquired double-time-phase remote sensing image data of the planet to-be-detected region into a planet remote sensing image change model which is established and trained in advance to obtain a surface terrain change detection result;
the planet remote sensing image change model extracts multilayer features of double time phase images based on a twin network framework, comprehensively measures the difference features of different levels through a multi-level difference feature fusion structure, and realizes detection of terrain change.
Compared with the prior art, the invention has the advantages that:
1. according to the invention, a planet remote sensing image change detection model based on a twin network frame is innovatively designed, wherein a main network adopts a MobileViT model to extract multilayer characteristics of a double-temporal image, so that inductive bias advantages of CNN and global information modeling advantages of a visual Transformer can be effectively combined, and the local and global visual characteristic characterization capabilities are enhanced;
2. the invention designs a multi-level difference characteristic fusion structure MDCF, comprehensively measures the difference characteristics of different levels, and enhances the discrimination capability of a model on the difference change information of double time phase images;
3. the change detection model designed by the invention not only has high accuracy, but also has the advantages of less parameters and low calculation complexity, and has higher practical application value.
Drawings
FIG. 1 is a model training and prediction flow diagram of the present invention;
FIG. 2 is a block diagram of a planetary remote sensing image change model structure according to the present invention;
fig. 3 is a diagram illustrating the overall implementation steps of the method for detecting the surface topography change of the planet remote sensing image based on the deep learning.
Detailed Description
The image data oriented to the algorithm is planet remote sensing image data obtained by observing the same region at different time, mainly aiming at the registered double-time-phase image, and the data type can be a visible light image or a multispectral image and the like.
The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and examples.
Example 1
Aiming at the problems of complicated calculation steps and high complexity of the existing related research method, the embodiment 1 of the invention designs a planetary remote sensing image change detection method based on deep learning by taking scene level change detection in a planetary earth surface range as a target, and the whole technical flow chart is shown in figure 1 and mainly comprises the following three steps:
step 1: firstly, a neural network model is constructed, a transPCD structure of a planet remote sensing image change model designed by the invention is shown in figure 2, and the change detection model mainly comprises the following components:
1) The double-temporal image feature extraction module: the model adopts a basic twin network frame, adopts two MobileViT network structures sharing parameters to perform multi-level feature extraction of the double-time-phase image, and respectively selects and obtains features of each level, namely Stage1_ T1, stage2_ T1, stage3_ T1, stage1_ T2, stage2_ T2 and Stage3_ T2 of the double-time-phase image. The MobileViT network combines the advantages of a convolutional neural network and a transformer structure based on a self-attention mechanism, wherein the mv module utilizes a transformer operation to replace a local processing step in a standard convolution operation, the defect that the receptive field of a convolutional layer is limited is overcome, and local and global visual information can be effectively characterized by using fewer parameters. The backbone network portion of MobileViT is utilized here, eliminating the last pooling layer and classifier of the original network.
2) Multilevel feature differential fusion structure (MFDF): comprehensively measuring the difference characteristics of different levels so as to enhance the characteristic expression capability of the model, wherein the calculation process comprises the following steps:
(1) inputting Stage1_ T1-Stage 3_ T1 and Stage1_ T2-Stage 3_ T2 into the MFDF module;
(2) calculating absolute difference values of all levels of features:
Figure BDA0003852409590000051
(2) then Diff is mixed 1 And Diff 2 The cascade operation of the two-stage differential feature is carried out to obtain Fdf 1 And then Fdf 1 Sequentially inputting the 3 × 3 standard convolutional layer and the maximum pooling layer to obtain Fdf 2 Then the output is compared with Diff 3 Cascading to obtain Fdf 3 The MFDF module outputs;
3) A classification decision module: mixing Fdf 3 And inputting the average pooling layer and the full-connection layer in sequence to obtain a final output result.
Step 2: after the model is built, training and evaluating the model by respectively using the existing training set and verification set, wherein a cross entropy loss function is adopted as a loss function in the training process, and an Adam optimizer is selected as the optimizer. In the training stage, a training set is used for training a model, sample data is input into the model in batches, a cross entropy loss function is calculated, gradient back propagation is carried out, an Adam optimizer is used for updating the network weight, iterative updating is carried out until the precision on the verification set is not improved any more, and then the training is finished;
and step 3: after training is finished, in order to detect the effect of the TransPCD model designed by the invention, a model test result is analyzed on a test data set, a comparison experiment is carried out with the existing change detection method, and the adopted precision evaluation index is the detection accuracy.
As shown in fig. 3, the whole implementation process of the planet remote sensing image terrain and surface change detection algorithm comprises the following steps:
(1) Firstly, acquiring double-time-phase planet remote sensing image data;
(2) Inputting the double time phase image to be tested into the trained model to obtain a prediction result;
(3) After the prediction result of the neural network model is obtained, a powerful basis can be provided for further research and verification of subsequent planet scientists.
The results of experiments performed on the published Mars image data demonstrate the effectiveness of the algorithm herein. The experimental result shows that the average detection precision of the model provided by the method reaches 96.5%, the accuracy is higher than that of a related algorithm, the size of the model is 1.38MB, the calculation cost is 0.19GFlops, the comprehensive performance is better, and the efficient planet remote sensing image change detection is realized.
Example 2
The embodiment 2 of the invention provides a system for detecting the surface topography change of a planet remote sensing image based on deep learning, which is realized by adopting the method of the embodiment 1, and comprises the following steps: the detection output module and the planet remote sensing image change model; wherein the content of the first and second substances,
the detection output module is used for inputting the acquired double-time-phase remote sensing image data of the planet to-be-detected region into a planet remote sensing image change model which is established and trained in advance to obtain a surface terrain change detection result;
the planet remote sensing image change model extracts multilayer features of double time phase images based on a twin network framework, and comprehensively measures the difference features of different levels through a multi-level difference feature fusion structure to realize detection of terrain change.
Finally, it should be noted that the above embodiments are only used for illustrating the technical solutions of the present invention and are not limited. Although the present invention has been described in detail with reference to the embodiments, it will be understood by those skilled in the art that various changes may be made and equivalents may be substituted without departing from the spirit and scope of the invention as defined in the appended claims.

Claims (7)

1. A planet remote sensing image earth surface terrain change detection method based on deep learning comprises the following steps:
inputting the acquired double-time-phase planetary remote sensing image data of the area to be detected into a planetary remote sensing image change model which is established and trained in advance, and obtaining a surface terrain change detection result;
the planet remote sensing image change model extracts multilayer features of double time phase images based on a twin network frame, comprehensively measures the difference features of different levels through a multi-level difference feature fusion structure, and realizes detection of terrain change.
2. The planetary remote sensing image earth surface terrain change detection method based on deep learning of claim 1, characterized in that the input of the planetary remote sensing image change model is remote sensing image data after planets are registered in the same region to be detected at T1 time and T2 time, the types of the remote sensing image data comprise a visible light image and a multispectral image, and the remote sensing image data are output as a detection result of whether earth surface terrain changes; the planet remote sensing image change model comprises a double time phase image feature extraction module, a multi-stage feature difference fusion module and a classification decision module; wherein the content of the first and second substances,
the double-temporal image feature extraction module is used for extracting multilayer features of double-temporal images and enhancing the local and global visual feature characterization capability;
the multi-level feature difference fusion module is used for comprehensively measuring the difference features of different levels and enhancing the feature expression capability;
and the classification decision module is used for outputting a detection result, wherein a value of 0 indicates that the detection result is not changed, and a value of 1 indicates that the detection result is changed.
3. The method for detecting the terrain change of the earth surface of the planet remote sensing image based on the deep learning as claimed in claim 1, wherein the double time-phase image feature extraction module adopts a basic twin network framework and comprises two improved MobileViT networks with the same structure, the improved MobileViT networks take a MobileViT backbone network part as a feature extractor, a final pooling layer and a classifier of an original MobileViT are removed, and three feature positions are respectively set at a mv 1 module output, a mv 2 module output and an improved MobileViT network output of each improved MobileViT network, so that a first feature Stage1_ T1, a second feature Stage2_ T1 and a third feature Stage3_ T1 of the planet remote sensing image data to be detected at the time T1, and a first feature Stage1_ T, a second feature Stage2_ T2 and a third feature Stage3_ T2 of the planet remote sensing image data to be detected at the time T2 are obtained.
4. The method for detecting the terrain change of the earth surface of the planet remote sensing image based on the deep learning as claimed in claim 3, wherein the processing process of the multi-stage feature difference fusion module MFDF comprises the following steps:
respectively calculating absolute difference values of the features Stage1_ T1 and Stage1_ T2, stage2_ T1 and Stage2_ T2 and Stage3_ T1 and Stage3_ T2 to obtain corresponding first-Stage absolute difference value Diff 1 Second order absolute difference Diff 2 Third order absolute difference Diff 3 ；
Figure FDA0003852409580000021
Diff is mixed 1 And Diff 2 Performing cascade operation to obtain a first cascade output Fdf 1 And then Fdf 1 Sequentially inputting the 3 x 3 standard convolution layer and the maximum pooling layer to obtain a second cascade output Fdf 2 And Fdf 2 And Diff 3 Cascade to obtain the third cascade output Fdf 3 And output to the classification decision module.
5. The method for detecting the change of the earth surface and the terrain of the planet remote sensing image based on the deep learning as claimed in claim 2, wherein the classification decision module comprises an average pooling layer and a full connection layer.
6. The method for detecting the change of the earth surface and the terrain of the planet remote sensing image based on the deep learning as claimed in claim 2, characterized by further comprising a training step of a planet remote sensing image change model; the method specifically comprises the following steps:
establishing a training set by using the public Mars image data;
inputting sample data in a training set into the planetary remote sensing image change model in batches, calculating a cross entropy loss function, performing gradient back propagation, updating network weights by adopting an Adam optimizer, and performing iterative updating until the training requirements are met to obtain the trained planetary remote sensing image change model.
7. The utility model provides a planet remote sensing image earth's surface topography change detecting system based on degree of deep learning, its characterized in that, the system includes: the detection output module and the planet remote sensing image change model; wherein the content of the first and second substances,
the detection output module is used for inputting the acquired double-time-phase remote sensing image data of the planet to-be-detected region into a planet remote sensing image change model which is established and trained in advance to obtain a surface terrain change detection result;
the planet remote sensing image change model extracts multilayer features of double time phase images based on a twin network framework, comprehensively measures the difference features of different levels through a multi-level difference feature fusion structure, and realizes detection of terrain change.
CN202211136723.5A 2022-09-19 2022-09-19 Planetary remote sensing image surface terrain change detection method based on deep learning Pending CN115424145A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211136723.5A CN115424145A (en) 2022-09-19 2022-09-19 Planetary remote sensing image surface terrain change detection method based on deep learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211136723.5A CN115424145A (en) 2022-09-19 2022-09-19 Planetary remote sensing image surface terrain change detection method based on deep learning

Publications (1)

Publication Number Publication Date
CN115424145A true CN115424145A (en) 2022-12-02

Family

ID=84204654

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211136723.5A Pending CN115424145A (en) 2022-09-19 2022-09-19 Planetary remote sensing image surface terrain change detection method based on deep learning

Country Status (1)

Country Link
CN (1) CN115424145A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116310851A (en) * 2023-05-26 2023-06-23 中国科学院空天信息创新研究院 Remote sensing image change detection method
CN116343052A (en) * 2023-05-30 2023-06-27 华东交通大学 Attention and multiscale-based dual-temporal remote sensing image change detection network
CN117173579A (en) * 2023-11-02 2023-12-05 山东科技大学 Image change detection method based on fusion of inherent features and multistage features
CN117409264A (en) * 2023-12-16 2024-01-16 武汉理工大学 Multi-sensor data fusion robot terrain sensing method based on transformer

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116310851A (en) * 2023-05-26 2023-06-23 中国科学院空天信息创新研究院 Remote sensing image change detection method
CN116310851B (en) * 2023-05-26 2023-08-15 中国科学院空天信息创新研究院 Remote sensing image change detection method
CN116343052A (en) * 2023-05-30 2023-06-27 华东交通大学 Attention and multiscale-based dual-temporal remote sensing image change detection network
CN117173579A (en) * 2023-11-02 2023-12-05 山东科技大学 Image change detection method based on fusion of inherent features and multistage features
CN117173579B (en) * 2023-11-02 2024-01-26 山东科技大学 Image change detection method based on fusion of inherent features and multistage features
CN117409264A (en) * 2023-12-16 2024-01-16 武汉理工大学 Multi-sensor data fusion robot terrain sensing method based on transformer
CN117409264B (en) * 2023-12-16 2024-03-08 武汉理工大学 Multi-sensor data fusion robot terrain sensing method based on transformer

Similar Documents

Publication Publication Date Title
CN115424145A (en) Planetary remote sensing image surface terrain change detection method based on deep learning
CN111126202B (en) Optical remote sensing image target detection method based on void feature pyramid network
CN108846835B (en) Image change detection method based on depth separable convolutional network
CN107563433B (en) Infrared small target detection method based on convolutional neural network
CN108596108B (en) Aerial remote sensing image change detection method based on triple semantic relation learning
CN110163213B (en) Remote sensing image segmentation method based on disparity map and multi-scale depth network model
CN111161224A (en) Casting internal defect grading evaluation system and method based on deep learning
CN114564982A (en) Automatic identification method for radar signal modulation type
Chen et al. Agricultural remote sensing image cultivated land extraction technology based on deep learning
CN114037891A (en) High-resolution remote sensing image building extraction method and device based on U-shaped attention control network
CN113989261A (en) Unmanned aerial vehicle visual angle infrared image photovoltaic panel boundary segmentation method based on Unet improvement
CN108960413A (en) A kind of depth convolutional neural networks method applied to screw surface defects detection
CN109671055A (en) Pulmonary nodule detection method and device
Holail et al. Afde-net: Building change detection using attention-based feature differential enhancement for satellite imagery
CN113610024B (en) Multi-strategy deep learning remote sensing image small target detection method
CN112818777B (en) Remote sensing image target detection method based on dense connection and feature enhancement
CN113095479B (en) Multi-scale attention mechanism-based extraction method for ice underlying structure
CN117830835A (en) Satellite remote sensing image segmentation method based on deep learning
CN115456957B (en) Method for detecting change of remote sensing image by full-scale feature aggregation
CN105654070A (en) Low-resolution face recognition method
CN113657290B (en) Snail collection and fine classification recognition system
CN116109813A (en) Anchor hole drilling identification method, system, electronic equipment and medium
CN116109826A (en) Road crack detection method
CN116189008A (en) Remote sensing image change detection method based on fixed point number quantification
CN115690597A (en) Remote sensing image urban ground feature change detection method based on depth background difference

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination